Chlorine and UV disinfection of ampicillin-resistant and trimethoprim-resistant Escherichia coli

Michael R. Templeton1*, Francine Oddy1, Wing-kit Leung2, and Michael Rogers1

1Department of Civil and Environmental Engineering, Imperial College London

South Kensington campus, London, United Kingdom SW7 2AZ

2Centre for Environmental Policy, Faculty of Natural Sciences, Imperial College London

South Kensington campus, London, United Kingdom SW7 2AZ

*Corresponding Author: Department of Civil and Environmental Engineering, Imperial College London, South Kensington campus, London, United Kingdom SW7 2AZ. Tel: +44 (0)207 594 6099; Fax: +44 (0)207 594 6124; Email: .

Word Count: 5,300 (counting Figures and Tables as 250-word equivalents each)

Submitted for publication in the Journal of Environmental Engineering and Science

Submitted 24 November 2008

Re-submitted with revisions 26 February 2009


Abstract

This bench-scale study investigated whether strains of Escherichia coli that are resistant to two common types of antibiotics, ampicillin and trimethoprim, possess increased resistance to two common disinfectants in water and wastewater treatment, free chlorine and ultraviolet disinfection, relative to an antibiotic-sensitive strain of E. coli isolated from sewage sludge. Trimethoprim-resistant E. coli was slightly more resistant to chlorine than the antibiotic-sensitive isolate and the ampicillin-resistant E. coli under the study conditions (95% confidence) however this difference would not be important under normal chlorination conditions applied in practice. There were no statistically significant differences between the ultraviolet dose-response profiles of the antibiotic-resistant and antibiotic-sensitive E. coli strains over the ultraviolet dose range tested.

Key Words: antibiotic-resistant bacteria, E. coli, UV disinfection, chlorine, water treatment, wastewater treatment.


Introduction

It has been observed that during water and wastewater treatment there is sometimes a greater proportion of antibiotic-resistant bacteria in treated water and wastewater effluents compared to the proportion in untreated water and wastewater, suggesting that treatment may somehow select for antibiotic-resistant survivors (Armstrong et al., 1981; Armstrong et al., 1982; Meckes 1982; Iwane 2001; Kim and Aga 2007). Several explanations have been proposed for this observation, such as a special phenotypic tolerance to chlorine (Earnhardt, 1981; Seyfried and Fraser, 1980) or protective capsule formation by antibiotic-resistant organisms (Reilly and Kippin, 1981). On the other hand, other studies have reported no effect or inconsistent effects of treatment on the selective survival of antibiotic-resistant bacteria post-treatment. For example, Bell et al. (1983) observed a decrease in antibiotic-resistant faecal coliform bacteria in short-term retention lagoons and mechanical treatment plants but an increase in these bacteria in long-term retention lagoons. Mezrioui and Baleux (1994) reported an increase in the percentage of antibiotic-resistant Escherichia coli during treatment in an aerobic lagoon, but no change in the percentage before and after an activated sludge process. Iwane et al. (2001) observed an increase in the percentage of ampicillin-resistant coliform bacteria after conventional wastewater treatment but a decrease in tetracycline-resistant bacteria. Kim et al. (2006) demonstrated an influence of changes in activated sludge process operating conditions on the fate of antibiotic-resistant bacteria in wastewater treatment.

The impact of disinfectants on antibiotic-resistant organisms has not been widely researched. Chlorination and ultraviolet (UV) disinfection are two of the most common disinfection processes currently used in water and wastewater treatment globally. Both have been proven to be effective against a range of waterborne pathogens, including E. coli, which is commonly used as an indicator organism (Baumann and Ludwig, 1962; Wilson et al., 1992; USEPA, 1999). The bactericidal effect of free chlorine is by non-specific oxidative interactions and affects a range of sub-cellular compounds and metabolic processes (Albrich and Hurst, 1982). It affects membrane permeability (Venkobachar et al., 1977), inhibits transport (Barrette et al., 1989), fragments proteins (Thomas, 1979), and reacts with nucleotides (Dennis et al., 1979). In addition, enzymes such as sulfhydryl enzymes are inactivated by chlorine (Knox et al., 1948), ATP production is inhibited by the oxidation of ATP-synthase (Barrette et al., 1989), and hypochlorite ions denature nucleic acids (Wlodkowski and Rosenkranz, 1975; Stewart and Olson, 1996). Inactivation of microorganisms by UV light is by an entirely different mechanism and involves the absorption of photons at wavelengths between 200 and 300 nm (especially 250-260 nm) by cellular components. Specifically, this absorption inhibits DNA/RNA transcription and replication due to pyrimidine dimerisation, most commonly the formation of thymine-thymine dimers (Patrick and Rahn, 1976; Durbeej and Ericksson, 2003; Hartl and Jones, 2005). Given that some previous studies have shown the potential for water and wastewater treatment to select for the survival of antibiotic-resistant species over antibiotic-sensitive species, it is important to investigate whether antibiotic-resistant bacteria may be more resistant to disinfection processes.

This study considered organisms with resistance to two antibiotics, ampicillin and trimethoprim. These antibiotics are both commonly used to combat infections caused by E. coli. Additionally, ampicillin and trimethoprim represent two important classes of anti-bacterial compounds, β-lactams and dihydrofolate reductase inhibitors, respectively, which operate against bacteria via different mechanisms.

Ampicillin (Figure 1(a)) belongs to the penicillin group of bactericidal β-lactam antibiotics, which are effective against Gram-positive bacteria and some Gram-negative bacteria such as E. coli. Following cell incursion, β-lactam antibiotics generally induce cell lysis by competitive inhibition of transpeptidase enzymes (also known as penicillin-binding proteins, PBPs) that are used during cell wall synthesis (Malouin and Bryan, 1986).

There are two main modes of resistance to ampicillin. The first involves hydrolysis of the antibiotic β-lactam ring by the bacterial enzyme β-lactamase, rendering ampicillin unable to bind to PBPs (Godfrey et al., 1981). β-Lactamase is often present on bacterial chromosomes or may otherwise be acquired by plasmid transfer. The second resistance mechanism involves the cell using altered PBPs for which ampicillin lacks the functionality to affect association (Godfrey et al., 1981).

Trimethoprim (Figure 1(b)) is a bacteriostatic antibiotic belonging to the diaminopyrimidine group of compounds. It acts by binding to the bacterial enzyme dihydrofolate reductase that is used to synthesise tetrahydrofolic acid, an essential precursor in the synthesis of the DNA nucleotide thymidine. The action of trimethoprim therefore starves bacteria of nucleotides required for DNA replication, greatly reducing the growth and reproduction potential of bacterial cells (Skold, 2001).

The primary resistance pathway to trimethoprim is through plasmid-mediated alteration of the target enzyme dihydrofolate reductase, with over twenty different genes known to code for this property (Skold, 2001). The circumvention of trimethoprim activity has also been observed via a bypass mechanism, mitigating the need for tetrahydrofolic acid in bacterial cell growth through the alternative use of folinic acid (Koneman et al., 2005).

The objective of this study was to determine whether these modes of bacterial resistance to ampicillin and trimethoprim also impart strains of Escherichia coli with increased resistance to chlorination and UV disinfection when compared to an antibiotic-sensitive E. coli isolate. Given that antibiotic-resistance mechanisms involve the alteration of cellular interactions with external antibiotics molecules, it was hypothesised that an impact on the penetration of chlorine molecules might be observed, whereas there was not expected to be an effect on the physical anti-microbial action of UV light.

Materials and Methods

Bacteria cultures and growth

Ampicillin-resistant E. coli 145 and trimethoprim-resistant E. coli 018 were obtained from a colleague at the Queen Mary’s School of Medicine and Dentistry in London. A wild strain of E. coli was also isolated from sewage sludge for comparison. The sludge-isolated E. coli was shown to be antibiotic-sensitive by failing to grow on tryptone soy broth agar (TSA, 37 g/L) media when trimethoprim (20 mg/L) or ampicillin (25 mg/L) was added (Table 1).

E. coli was grown overnight (18 hours) in 50 mL of tryptone soy broth (TSB, 30 g/L), with additional trimethoprim (20 mg/L) or ampicillin (25 mg/L) added when considering E. coli 018 and 145, respectively, in 250 mL conical flasks shaken at 100 rpm at 20 ˚C.

Chlorination protocol

A chlorine stock solution with a concentration of 100mg/L as Cl2 was prepared by dosing 0.8 mL sodium hypochlorite in 500 mL deionised water. This chlorine stock solution was stored in an aluminum-foil covered, glass-capped flask in the dark. The concentration was checked periodically by standard DPD titration (APHA, 2005).

E. coli in the overnight TSB growth medium was harvested and washed three times with sterile phosphate buffer saline (PBS) pH 7 buffer and centrifuged at 3000 rpm for 10 minutes in sterile centrifuge tubes. This washing step was included to reduce the chlorine demand of the bacterial cultures. The harvested E. coli were then re-suspended in 50 mL of sterile buffer. The microbial suspension was inoculated into 200 ml sterile PBS (pH 7) to obtain a bacterial concentration on the order of 106-107 colony-forming units per millilitre (cfu/mL). The suspension was then stirred with a magnetic stir bar at 20ºC.

A time zero sample for E. coli enumeration was taken and then a 3.0 ml aliquot of the chlorine stock solution was added to achieve an initial chlorine dose of 1.25 mg/L as Cl2, which was found beforehand to be the initial chlorine dose that would result in a minimum 0.05 mg/L free chlorine residual at the end of the 17-minute chlorination trials (i.e. by taking into account the chlorine demand of the bacterial suspension itself). At 2, 5, 9, 13, and 17 minutes, 10 mL samples were removed using a sterile pipette for DPD titration to determine total and free chlorine residuals.

At 0, 1, 4, 7, 11, and 15 minutes, 10 mL samples were removed using a sterile pipette and transferred into sterile culture tubes containing 2 mL of concentrated sodium thiosulphate and shook on a vortex mixer. These samples were used for E. coli enumeration. Serial dilutions were made for each sample, with the aim of obtaining 20–80 colonies per plate. This was achieved by transferring 1ml of sample sequentially into sterile sampling tubes containing 9mL of PBS at pH 7, with shaking on a vortex mixer. These dilutions were conducted in duplicate. The serial dilutions were then poured through a 0.45-µm pore, 47 mm diameter sterile membrane filter under vacuum. Each filter was then transferred onto a 55-mm diameter TSA plate (containing trimethoprim, ampicillin or no antibiotics, as appropriate), inverted and incubated at 35 °C for 20 hours (APHA, 2005). This was repeated for the duplicate serial dilution series. Colonies were counted the following day.

Each chlorination trial was repeated under identical conditions in triplicate.

UV disinfection protocol

The PBS solution inoculated with E. coli was prepared as described above for the chlorination trials. Samples (20 ml) of the solution were transferred into 90-mm diameter Petri dishes using sterile pipettes. Samples were exposed under a UV collimated beam to a range of UV doses (0, 1, 3, and 5 mJ/cm2) by varying the exposure time under the beam. The UV doses considered in this study are considerably lower than the doses used in treatment practice (e.g. 20-200 mJ/cm2) but were selected in order to be able to examine the potential differences in the UV dose-response profiles for the three strains of E. coli, which required countable surviving organisms post-UV exposure. Standard methods for UV dose measurement and calculations for collimated beam experiments were followed, as described elsewhere (Bolton and Linden, 2003). An IL1700 radiometer equipped with SED240 sensor (International Light, Peabody, MA, USA) was used to measure the UV intensity across the exposure surface, taking readings every 0.5 cm (Bolton and Linden, 2003). The reading at the centre of the exposure surface was 0.247 mW/cm2. A 1-cm magnetic stir bar was used to gently stir the samples during the UV exposures. After UV exposure the samples were serially diluted with the aim of obtaining 20-200 colonies per plate. The exposed sample (0.1 ml) was then transferred onto TSA plates, containing trimethoprim (20 mg/L), ampicillin (25 mg/L), or no antibiotic when using E. coli 018, E. coli 145, or the sludge-isolated antibiotic-sensitive E. coli, respectively, spread, then left on the bench for 20 minutes to absorb and dry. The plates were then inverted and incubated for 14-16 hours before counting colony forming units. Each UV dose exposure and serial dilution was plated in duplicate and the entire experiment was repeated in triplicate.

Bacteriophage MS2, a standard surrogate organism in UV disinfection studies with a well-known UV dose-response profile (Wilson et al., 1992; USEPA, 2006) was used as a benchmark for confirmation of the UV doses that were applied by the collimated beam. The methods for growth and enumeration of phage MS2 have been described elsewhere (USEPA, 2001; Templeton et al., 2005). A UV dose of 40 mJ/cm2 should result in approximately 2.5-log inactivation of phage MS2 in low UV absorbance, particle-free water (Templeton et al., 2005). In this study, the inactivation of phage MS2 by 40 mJ/cm2 was 2.6-log on average based on three replicate exposures, which validated the UV dose exposure method.

For both the chlorination and UV disinfection trials, statistical significance was assessed by t-test comparisons of data sets at a 95% confidence level (n = 3).

Results and Discussion

Chlorination results

The log inactivation and chlorine decay results for the chlorination trials are summarised in Figure 2, where N0 represents the initial colony counts and N represents the colony counts at the indicated sampling time. The bacterial suspensions exerted a high chlorine demand, resulting in a decrease in chlorine residual to 0.05-0.30 mg/L after only 1-2 minutes but then it stabilised for the remainder of the trial. Log inactivation followed a similar pattern, with high levels of bacterial inactivation early on and then a more gradual trend over the remainder of the trial. Tailing of dose-response curves is a common observation in disinfection studies and has been attributed to factors such as clumping of cellular matter or phenotypic disinfection-resistance of sub-populations (Cerf, 1977; Stewart and Olsen, 1996).

The log inactivation results for the three strains of E. coli are expressed relative to the applied ‘Ct’ (i.e. the integrated product of chlorine residual, ‘C’, and contact time, ‘t’) in Figure 3. The trimethoprim-resistant strain of E. coli 018 was more resistant to chlorine than the antibiotic-sensitive E. coli isolate, and the ampicillin-resistant E. coli 145 was less resistant to chlorine than the E. coli isolate, both to statistically significant degrees (95% confidence). However, it should be noted that the Ct values considered in this experiment (0 to 5 mg∙min/L) are very low compared to what is typically applied in treatment practice (USEPA, 1999). For example, the Ct required for 3-log inactivation of Giardia lamblia with 1 mg/L free chlorine at pH 7.0 and 20 ºC is 56 mg∙min/L (Malcolm Pirnie and HDR Engineering, 1991). In this study, greater than 4-log inactivation was achieved for all of the strains of E. coli at a Ct of only 2.5 mg∙min/L. Therefore, the differences in chlorine-resistance observed in this study are not likely to be relevant under normal chlorination conditions.